Wilson Bull., 105(3), 1993, pp. 399-413 POPULATION STRUCTURE AND GENE FLOW IN THE CHIPPING SPARROW AND A HYPOTHESIS FOR EVOLUTION IN THE GENUS SPIZELLA ROBERT M. ZINK’ AND DONNA L. DITTMANN’ AssTnAcr.-We studied restriction site variation in the mitochondrial DNA (mtDNA) of 55 Chipping Sparrows (Spizella passerina) taken from widely dispersed points in their breeding range. A total of 2 1 haplotypes was observed, and on average, individuals differed little in percent haplotype divergence (0.12%). There was no detectable geographic variation in haplotypes, despite the sampling of three named subspecies. Single-generation dispersal distance was estimated from the mtDNA data at 5.4 km. One phylogenetic hypothesis for six species in the genus (excluding Timberline Sparrow [S. taverneri] and Worthen’s Sparrow [S. worthenl]) suggested that Black-chinned Sparrow (S. atroguluris) and Field Sparrow (S. pusilla) were sister taxa, followed in sequence by Chipping, Brewer’s and Clay-colored sparrows (S. pallida), with the American Tree Sparrow (S. arboreu) most distant. Another hypothesis grouped Chipping and Brewer’s sparrows. Received 24 Aug. 1992, accepted 3 Feb. 1993. An important step in the evolutionary process is the origin of geographic patterns of genetic variation. Patterns of genetic variation form the basis for inferences about evolutionary processes that lead to the origin and maintenance of geographic variation and speciation. In the last 20 years, a number of molecular techniques, including protein electrophoresis and restriction site analysis of mitochondrial DNA (mtDNA; Avise et al. 1987), have been used to describe genetic variation within and among avian populations. For north temperate birds, protein electrophoresis revealed few alleles that varied geographically (Barrowclough 1983). From this result, many researchers inferred that avian populations were essentially panmictic, and that gene flow was substantial (Barrowclough 1983, Barrowclough and Johnson 1988). Studies of mtDNA opened a new era because of several advantageous aspects of mtDNA evolution (Moritz et al. 1987). In particular, mtDNA appears to evolve faster than most types of nuclear DNA (including that which encodes allozymes), providing an enhanced probability of detecting geographic variation. Indeed, mtDNA studies reveal an array of population structures (Ball et al. 1988, Avise and Nelson 1989, Zink 1991, Avise and Ball 1991, Moore et al. 1991, Shields and Wilson 1987a, Ball and Avise 1992, Zink and Dittmann, in press). For instance, Avise and Nelson (1989) found two geographic groups of mtDNA lineages in the Seaside Sparrow (Ammodramus maritimus). Zink (199 1, I Museum of Natural Science, Louisiana State Univ., Baton Rouge, Louisiana 70803. (Present address RMZ: Bell Museum of Natural History, Ecology Bldg., Univ. of Minnesota, St. Paul, Minnesota 55108.) 399 400 THE WILSON BULLETIN l Vol. 105, No. 3, September 1993 in press) found significant mtDNA variation in the Fox Sparrow (Pus- serella iliaca), whereas in an allozyme survey no geographic differentiation was detected (Zink 1986). In the Song Sparrow (Melospiza melodia), which exhibits extensive geographic variation in morphology, there was consid- erable mtDNA variation but essentially no geographic structure (Zink and Dittmann, in press). To date, mtDNA studies of geographic variation in birds have tended to involve species that exhibit geographic variation in morphology. Con- sequently, we know little about the genetic structure of species that exhibit little or no geographic variation in external phenotype. Because it is often assumed that geographic variation in external phenotypes reflects under- lying genetic structure, it is appropriate to examine the geographically uniform species. Such comparisons will clarify the relationship between morphological and genetic variation. In this paper, we describe a survey of mtDNA restriction site variation in the Chipping Sparrow (Spizella passerina), a widespread migratory North American passerine bird (Fig. 1). Taxonomic variation in the Chipping Sparrow in North America (north of Mexico) is limited to three weakly demarcated subspecies (AOU 1957). Our samples were taken from regions as far apart as North Carolina and the Yukon. We estimate the degree of population subdivision, level of gene flow, and the correspondence of mtDNA variation and subspecies limits. To put our study in a phylogenetic perspective, we also provide an estimate of the relationships of species in the genus, with the exception of the Timberline Sparrow (S. taverneri) and Worthens’ Sparrow (S. war- theni), species for which we lacked tissue samples. METHODS Samples (Table 1) were collected during the breeding season from 13 sites (Fig. 1). From each of the 55 specimens, samples of liver, heart, and pectoral muscle were frozen in liquid nitrogen. Upon return to the laboratory, mtDNA was purified from tissues following pro- cedures in Lansman et al. (198 1) and Dowling et al. (1990). Purified mtDNA from each individual was digested with 20 restriction endonucleases (see Appendix I), end-labeled with 95S, and electrophoresed in 0.7% to 1.1% agarose gels. Fragments were visualized with autoradiography. A letter was assigned to each distinct restriction fragment profile. The combination of an individuals’ letters for all endonucleases constitutes its composite hap- lotype. Each individual was scored for the presence or absence of each restriction fragment for each endonuclease. Fragment profiles produced by each endonuclease differed from each other by one or two restriction sites, making inference of the presence or absence of restriction sites straightforward. The matrix of restriction sites was used to estimate the percent nu- cleotide divergence (p) among haplotypes following Nei and Li (1979). A computer program written by J. E. Neigel was used to estimate single generation dispersal distance (Neigel et al. 199 1). The method requires a phenogram based on mtDNA genetic distances (p-values), generation length (we assumed that two years was reasonable), geographic distance between each pair of localities, and an estimate of mtDNA evolutionary rate, for which we used one percentage per lineage per million years (Shields and Wilson 1987b). The procedure involves Zink and Dittmann l MTDNA STUDIES OF SPIZELLA 401 FIG. 1. Approximate breeding distribution of the three subspecies of the Chipping Spar- row, and location and codes of collecting sites (see Table 1). dividing the phenogram into 10 equal sections based on level of divergence (beginning at p = O.O), and estimating the dispersal distance for each section. It is expected (J. E. Neigel, pers. comm.) that the youngest lineage classes (i.e., the first few sections in the phenogram) will provide the best estimate of dispersal distance because they likely have not been limited or confined by geographic barriers to gene flow (therefore the relationship between genetic and geographic distance is monotonic). The matrix of restriction sites was analyzed by the computer program HENNIG86 (Fanis 1989) to infer a phylogenetic network using the principle of maximum parsimony. The matrix of p-values was analyzed using the FITCH routine in the computer program PHYLIP (Felsenstein 1987) which produces a distance tree following the method of Fitch and Margoliash (1967). Also, a UPGMA phenogram was produced from the matrix ofp-values (Sneath and Sokal 1973). For the phylogenetic analysis of species, only the pattern of restriction fragments (see Appendix 2) was analyzed because we lacked financial resources to map restriction sites; see Zink and Avise (1990) for justi- fication. One individual per species was used. Both HENNIG86 and PHYLIP were used to 402 THE WILSON BULLETIN l Vol. 105, No. 3, September 1993 TABLE 1 DISTRIBUTION OF SAMPLES (SEE FIG. 1) AND COMPOSITE HAPLOTYPES FOUND AT EACH SAMPLE SITES Geographicsite (code) N Haplotype(9 1. North Carolina (NC) 1 1 2. Wisconsin (WI) 2 1 (2) 3. Minnesota (MN) 2 1 (2) 4. Louisiana (LA) 3 1 (2) 2 5. Manitoba (MA) 8 l(4), 3, 7, 1521 6. Saskatchewan (SA) 7 1 (3) 4, 5, 16, 17 7. Alberta (AL) 3 1367 8. Yukon (YU) 2 1 (2) 9. Northern British Columbia (NB) 7 l(4), 8, 18, 19 10. Central British Columbia (CB) 3 1,7, 10 11. Southeastern British Columbia (SB) 5 1 (3) 20, 21 12. Washington (WA) 3 1 (2) 9 13. Southern California (CA) 9 1 (5) 11, 12, 13, 14 a Numbersof individualswith particularhaplotype in parentheses. Samples l-4 are from .S. p. passerim. 5-1 I from S. p. boreophila. and 12-l 3 from 5’. p. arrzonae. analyze the fragment matrix as described above. We also used the bootstrap procedure (Felsenstein 1985) in PHYLIP to evaluate the strength of the phylogenetic pattern obtained; because not all restriction fragments are independent, we do not interpret bootstrap values statistically but rather as a description of the data (Zink and Avise 1990). RESULTS Chipping Sparrow. -Based on restriction fragment profiles that yielded mtDNA fragments between 1000 and 8000 base pairs, we estimated the size of the mtDNA genome at 16,500. Five restriction endonucleases produced only a single pattern: Ava I, Kpn I, Nci I, Sac II, and Stu I. The other 15 endonucleases produced from two to four patterns (Appendix I). A total of 14 1 restriction sites was found. A total of 2 1 haplotypes was observed among the 55 Chipping Sparrows studied. Apart from haplotype 1, only haplotype 7 was identified in more than one individual (Table 1). Haplotype 1 occurred in every geographic sample and in 32 of the 55 individuals. Most haplotypes differed by one or two restriction sites. The average p-value among the 21 haplotypes was 0.28% -t 0.12 (SD; N = 210), and the range was from 0.06 to 0.62%. Among the 55 individuals, the average p-value was 0.12% k 0.12 (SD), and the range was from zero to 0.62% (N = 1485). The distribution (Fig. 2) ofp-values among all pairs of haplotypes reveals that most individuals have very similar mtDNA genomes. The topology of the UPGMA phenogram depends on the input order of haplotypes (because of identical values in the distance matrix), Zink and Dittmann l MTDNA STUDIES OF SPIZELLA 403 0 1 2 3 4 5 6 7 Percent HaplotypeDivergence FIG.